U.S. patent application number 10/885259 was filed with the patent office on 2006-01-12 for digital photography with flash/no flash extension.
This patent application is currently assigned to Microsoft Corporation. Invention is credited to Maneesh Agrawala, Michael F. Cohen, Hugues Hoppe, Georg F. Petschnigg, Richard Szeliski.
Application Number | 20060008171 10/885259 |
Document ID | / |
Family ID | 35116074 |
Filed Date | 2006-01-12 |
United States Patent
Application |
20060008171 |
Kind Code |
A1 |
Petschnigg; Georg F. ; et
al. |
January 12, 2006 |
Digital photography with flash/no flash extension
Abstract
A system and method for improving digital flash photographs. The
present invention is a technique that significantly improves
low-light imaging by giving the end-user all the advantages of
flash photography without producing the jarring look. The invention
uses an image pair--one taken with flash the other without--to
remove noise from the ambient image, sharpen the ambient image
using detail from the flash image, correct for color, and remove
red-eye.
Inventors: |
Petschnigg; Georg F.;
(Seattle, WA) ; Szeliski; Richard; (Bellevue,
WA) ; Cohen; Michael F.; (Seattle, WA) ;
Hoppe; Hugues; (Redmond, WA) ; Agrawala; Maneesh;
(Seattle, WA) |
Correspondence
Address: |
AMIN & TUROCY, LLP
24TH FLOOR, NATIONAL CITY CENTER
1900 EAST NINTH STREET
CLEVELAND
OH
44114
US
|
Assignee: |
Microsoft Corporation
Redmond
WA
98052
|
Family ID: |
35116074 |
Appl. No.: |
10/885259 |
Filed: |
July 6, 2004 |
Current U.S.
Class: |
382/254 |
Current CPC
Class: |
G06T 5/50 20130101 |
Class at
Publication: |
382/254 |
International
Class: |
G06K 9/40 20060101
G06K009/40 |
Claims
1. A system that facilitates generation of digital images,
comprising: a component that receives a plurality of digital images
of an object, wherein at least two of the digital images have
differing illumination components; and an image generation
component that merges a subset of the received images to generate a
new image of the object.
2. The system of claim 1, wherein one of the at least two digital
images is taken with flash lighting.
3. The system of claim 1, wherein one of the at least two digital
images is taken with flash lighting of higher intensity than a
flash lighting for another image.
4. The system of claim 1, wherein a subset of the plurality of
images are taken consecutively within a predetermined time
frame.
5. A digital camera comprising the system of claim 1.
6. The system of claim 1 further comprising a comparison component
that identifies an image artifact by analyzing the at least two of
the digital images.
7. The system of claim 6, wherein the image artifact is at least
one of red-eye, a shadow, and noise.
8. The system of claim 1, wherein the image generation component
merges high frequency components of one image with low frequency
components of another image in generating the new image.
9. The system of claim 8, further comprising a cost component that
applies a cost function to the frequency components to mitigate
combining disagreeing frequency components.
10. The system of claim 8, wherein at least one of a bi-lateral
filter and a joint bilateral filter is employed.
11. The system of claim 1, wherein the at least two of the digital
images comprise a flash and no-flash pair taken in rapid
succession
12. The system of claim 1, wherein color in one of the digital
images is continuously corrected.
13. The system of claim 1, wherein the image generation component
enhances one image by adding detail from another image.
14. A portable wireless communications device comprising the system
of claim 1.
15. A computer readable medium having stored thereon computer
executable instructions for carrying out the system of claim 1.
16. A system that facilitates generation of digital images,
comprising: a component that receives a plurality of digital images
of an object, wherein at least two of the plurality of digital
images have differing spatial frequency components; and an image
generation component that merges a subset of the received plurality
of digital images to generate a new image of the object.
17. The system of claim 16, wherein the-at least two of the
plurality of digital images includes a flash image that includes
high frequency spatial components and a no-flash image that
includes low frequency spatial components.
18. A handheld computing device that employs the system of claim
16.
19. A computer-readable medium having computer-executable
instructions for performing a method for facilitating the
generation of digital images, the method comprising: receiving a
flash image and a no-flash image of an object; denoising the
no-flash image using the flash image; transferring a detail of the
flash image to the no-flash image; and outputting a new image of
the object based on the flash image and the no-flash image.
20. The method of claim 19, further comprising performing
white-balancing by using flash color of the flash image to
white-balance the no-flash image.
21. The method of claim 19, further comprising performing red-eye
correction by processing pupil color between the flash image and
the no-flash image.
22. The method of claim 19, further comprising detecting flash
shadows and specular regions.
23. The method of claim 19, the act of denoising further comprises
estimating high-frequency information using the flash image.
24. The method of claim 19, the act of transferring further
comprises: detecting at least one of regions of shadows and
specularities; and generating a mask that identifies the
regions.
25. The method of claim 24, further comprising interactively
adjusting a threshold value related to the shadows.
26. A method of facilitating the generation of a digital image,
comprising: receiving a flash image and a no-flash image of an
object; denoising the no-flash image using the flash image; and
outputting a new image based on the flash image and the no-flash
image.
27. The method of 26, further comprising merging high frequency
components of the flash image with low frequency components of the
no-flash image.
28. The method of 26, further comprising applying a cost function
that prevents combining of disagreeing frequency components.
29. The method of 26, the act of denoising further comprises:
filtering the no-flash image using a bilateral filter that averages
pixels that are at least one of spatially-near and have similar
intensity values; and estimating high-frequency components of the
flash image using a joint bilateral filter.
30. The method of 26, further comprising performing white-balancing
by estimating ambient color illumination.
31. The method of claim 30, further comprising analyzing the
estimated color illumination at a plurality of pixels of the
no-flash image.
32. The method of 26, further comprising at least one of the acts
of: computing a detail layer from the flash image according to a
ratio; and detecting shadows and specularities in the flash
image.
33. A method of facilitating the generation of a digital image,
comprising: receiving a flash image and a no-flash image of an
object; transferring a detail of the flash image to the no-flash
image; and outputting a new image of the object based on the flash
image and the no-flash image.
34. The method of claim 33, the act of receiving comprises
obtaining the flash image and the no-flash image successively in
less than 1/30 second.
35. The method of claim 33, further comprising computing a detail
layer of the flash image according to a ratio, which ratio is
computed on at least one RGB channel.
36. The method of claim 35, the ratio is independent of signal
magnitude and signal reflectance.
37. The method of claim 33, further comprising estimating a mask
that identifies at least one of a shadow region and a specular
region.
38. The method of claim 37, further comprising generating the mask
by merging a mask of the shadow region and a mask of the specular
region.
39. The method of claim 33, further comprising controlling an
amount of the detail transferred using a filter.
40. A system that facilitates generation of a digital image,
comprising: means for receiving a flash image and a no-flash image
of an object; means for denoising the no-flash image; means for
transferring a detail of the flash image to the no-flash image;
means for white-balancing the no-flash image using the flash image;
means for adjusting flash intensity after the object is captured;
means for correcting for an artifact; and means for outputting a
new image based on the flash image and the no-flash image.
Description
TECHNICAL FIELD
[0001] This invention is related to digital photography, and more
specifically to mechanisms for improving the quality of digital
flash photographs.
BACKGROUND OF THE INVENTION
[0002] An important goal of photography is to capture and reproduce
the visual richness of a real environment. Lighting is an integral
aspect of this visual richness and often sets the mood or
atmosphere in the photograph. The subtlest nuances are often found
in low-light conditions. For example, the dim, orange hue of a
candlelit restaurant can evoke an intimate mood, while the pale
blue cast of moonlight can evoke a cool atmosphere of mystery.
[0003] When capturing the natural ambient illumination in such
low-light environments, photographers face a dilemma. One option is
to set a long exposure time so that the camera can collect enough
light to produce a visible image. However, camera shake or scene
motion during such long exposures will result in motion blur.
Another option is to open the aperture to let in more light.
However, this approach reduces depth of field and is limited by the
size of the lens. The third option is to increase the camera's
gain, which is controlled by the ISO setting. However, when
exposure times are short, the camera cannot capture enough light to
accurately estimate the color at each pixel, and thus visible image
noise increases significantly.
[0004] Flash photography was invented to circumvent these problems.
By adding artificial light to nearby objects in the scene, cameras
with flash can use shorter exposure times, smaller apertures, and
less sensor gain and still capture enough light to produce
relatively sharp, noise-free images. Brighter images have a greater
signal-to-noise ratio and can therefore resolve detail that would
be hidden in the noise in an image acquired under ambient
illumination. Moreover, the flash can enhance surface detail by
illuminating surfaces with a crisp point light source. Finally, if
one desires a white-balanced image, the known flash color greatly
simplifies this task.
[0005] As photographers know, however, the use of flash can also
have a negative impact on the lighting characteristics of the
environment. Objects near the camera are disproportionately
brightened, and the mood evoked by ambient illumination may be
destroyed. In addition, the flash may introduce unwanted artifacts
such as red eye, harsh shadows, and specularities, none of which
are part of the natural scene. Despite these drawbacks, many
amateur photographers use flash in low-light environments, and
consequently, these snapshots rarely depict the true ambient
illumination of such scenes.
SUMMARY OF THE INVENTION
[0006] The following presents a simplified summary of the invention
in order to provide a basic understanding of some aspects of the
invention. This summary is not an extensive overview of the
invention. It is not intended to identify key/critical elements of
the invention or to delineate the scope of the invention. Its sole
purpose is to present some concepts of the invention in a
simplified form as a prelude to the more detailed description that
is presented later.
[0007] The present invention disclosed and claimed herein, in one
aspect thereof, comprises a system and method for improving digital
flash photographs. Flash photography in general looks bad. The
present invention overcomes many of the drawbacks and shortcomings
of the prior art by providing a technique that significantly
improves low-light imaging by giving the end-user all the
advantages of flash photography without producing the jarring look.
In operation, the present invention uses an image pair--one taken
with flash the other, the ambient image, without--to remove noise
from the ambient image, sharpen the ambient image using detail from
the flash image, correct for color, and remove red-eye.
[0008] In one aspect thereof, the present invention uses the flash
image's better signal characteristics to drive the de-noising of
the ambient image.
[0009] In another aspect of the present invention, the present
invention uses the fact that the color exposed by the flash is
known to more robustly estimate the ambient illumination in the
non-flash image, to create a more natural looking ambient
image.
[0010] In yet another aspect thereof, a variety of applications are
provided that analyze and combine the strengths of such
flash/no-flash image pairs. These applications include denoising
and detail transfer (to merge the ambient qualities of the no-flash
image with the high-frequency flash detail), white-balancing (to
change the color tone of the ambient image), continuous flash (to
interactively adjust flash intensity), and red-eye removal (to
repair artifacts in the flash image).
[0011] In still another aspect of the present invention, manual
acquisition of the flash/no-flash pair is provided that is
relatively straightforward with current consumer digital
cameras.
[0012] To the accomplishment of the foregoing and related ends,
certain illustrative aspects of the invention are described herein
in connection with the following description and the annexed
drawings. These aspects are indicative, however, of but a few of
the various ways in which the principles of the invention can be
employed and the present invention is intended to include all such
aspects and their equivalents. Other advantages and novel features
of the invention will become apparent from the following detailed
description of the invention when considered in conjunction with
the drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] FIG. 1 illustrates a system that facilitates digital image
generation using flash/no-flash image pairs in accordance with the
present invention.
[0014] FIG. 2 illustrates a flow chart of one methodology for new
image generation in accordance with the present invention.
[0015] FIG. 3 illustrates a flow chart of one methodology of
acquisition in accordance with the present invention.
[0016] FIG. 4 illustrates a flow chart of one methodology for
denoising in accordance with the present invention.
[0017] FIG. 5 illustrates a flow chart of one methodology for
detail transfer in accordance with the present invention.
[0018] FIG. 6 illustrates an overview of a denoising algorithm, a
detail transfer algorithm, and flash artifact detection algorithm
that operate on a no-flash image and a flash image, in accordance
with the present invention.
[0019] FIG. 7a illustrates a close-up of a flash image and a
no-flash image of a Belgian tapestry.
[0020] FIG. 7b shows a denoised image via basic bilateral filtering
to preserve strong edges, but blurs away most of the threads.
[0021] FIG. 7c shows a denoised image using joint bilateral
filtering.
[0022] FIG. 7d shows the difference image between the basic and
joint bilateral filtered images.
[0023] FIG. 7e shows an example of a detail layer.
[0024] FIG. 7f shows an example of a detail transfer.
[0025] FIG. 8a shows a flash image and a no-flash image of an old
European lamp made of hay.
[0026] FIG. 8b illustrates a small section of the image that is
examined for final results.
[0027] FIG. 8c shows the flash version of the section of FIG.
8b
[0028] FIG. 8d shows the no-flash version of the section of FIG.
8b.
[0029] FIG. 8e shows the detail transfer with denoising, that
maintains the warm appearance, as well as the sharp detail.
[0030] FIG. 9a shows a no-flash image.
[0031] FIG. 9b shows the detail transfer with denoising of FIG.
9a.
[0032] FIG. 9c shows a long exposure image of the wine cave scene
(3.2 seconds at ISO 100) that is captured for comparison with
detail transfer with denoising result of FIG. 9b.
[0033] FIG. 10 illustrates a flow chart of one methodology for
flash shadow and specularity detection in accordance with the
present invention.
[0034] FIG. 11a illustrates flash and no-flash images.
[0035] FIG. 11b shows the detail transfer image for the no-flash
image of FIG. 11a.
[0036] FIG. 11c shows the detail transfer without Mask of the
section of the flash image, where shadows are outlined at the
arrows.
[0037] FIG. 11d shows the shadow and specularity mask.
[0038] FIG. 11e shows the detail transfer using the Mask of the
same image of FIG. 11c.
[0039] FIG. 11f shows a flash image of the small section indicated
in the no-flash image of FIG. 11a.
[0040] FIG. 11g shows the no-flash version of the same small
section of FIG. 11f.
[0041] FIG. 11h shows the detail transfer with denoising of the
images of FIG. 11f and FIG. 11g.
[0042] FIG. 12 illustrates a flow chart of one methodology for
white balancing in accordance with the present invention.
[0043] FIG. 13a shows an original no-flash image after denoising
and detail transfer, but which still shows a cast.
[0044] FIG. 13b shows the estimated ambient illumination colors and
the estimated overall scene ambience.
[0045] FIG. 13c shows that the white-balancing algorithm shifts the
colors and removes a certain coloring.
[0046] FIG. 14 illustrates a flow chart of one methodology for
continuous flash adjustment in accordance with the present
invention.
[0047] FIG. 15a shows an out-of-range extreme at the low end with a
setting at -0.5.
[0048] FIG. 15b shows the no-flash image at 0.0.
[0049] FIG. 15c shows an extrapolated image with a 0.33
setting.
[0050] FIG. 15d shows an extrapolated image with a 0.66
setting.
[0051] FIG. 15e shows the flash image at the 1.0 setting.
[0052] FIG. 15f shows an extrapolated image with an out-of-range
extreme at the high end at 1.5.
[0053] FIG. 16 illustrates a red-eye removal methodology in
accordance with the present invention.
[0054] FIG. 17 illustrates a block diagram of a small form factor
portable device that includes the image processing architecture of
the present invention.
[0055] FIG. 18 illustrates a block diagram of a computer operable
to execute the disclosed architecture.
[0056] FIG. 19 illustrates a schematic block diagram of an
exemplary computing environment in accordance with the present
invention.
DETAILED DESCRIPTION OF THE INVENTION
[0057] The present invention is now described with reference to the
drawings, wherein like reference numerals are used to refer to like
elements throughout. In the following description, for purposes of
explanation, numerous specific details are set forth in order to
provide a thorough understanding of the present invention. It may
be evident, however, that the present invention can be practiced
without these specific details. In other instances, well-known
structures and devices are shown in block diagram form in order to
facilitate describing the present invention.
[0058] As used in this application, the terms "component" and
"system" are intended to refer to a computer-related entity, either
hardware, a combination of hardware and software, software, or
software in execution. For example, a component can be, but is not
limited to being, a process running on a processor, a processor, an
object, an executable, a thread of execution, a program, and/or a
computer. By way of illustration, both an application running on a
server and the server can be a component. One or more components
can reside within a process and/or thread of execution, and a
component can be localized on one computer and/or distributed
between two or more computers.
[0059] The present invention overcomes many of the drawbacks and
shortcomings of the prior art by providing a technique that
significantly improves low-light imaging by giving the end-user all
the advantages of flash photography without producing the jarring
look. In operation, the present invention uses an image pair--one
taken with flash, the other without (the "ambient" image)--to
remove noise from the ambient image, sharpen the ambient image
using detail from the flash image, correct for color, and remove
red-eye. In one embodiment, the present invention uses the better
signal-to-noise characteristics of a flash image to drive the
denoising of an ambient (or no-flash) image. In a second
embodiment, the present invention uses the fact that color exposed
by the flash is known to more robustly estimate the ambient
illumination in the non-flash image, to create a more natural
looking ambient image.
[0060] Referring now to FIG. 1, there is illustrated a system 100
that facilitates digital image generation using flash/no-flash
image pairs in accordance with the present invention. The system
100 includes an input component 102 that receives a plurality of
digital images 104 (denoted INPUT IMAGE.sub.1, INPUT IMAGE.sub.2, .
. . , INPUT IMAGE.sub.N) generated of an object 106. At least two
of the digital images 104 have differing illumination components.
An image generation component 108 merges a subset of the input
images 104 to generate a new image 110 of the object 106. The image
generation component 108 merges high frequency components of one
input image with low frequency components of another input image in
generating the new image 110, thereby enhancing one image by adding
detail from another image.
[0061] The new image 110 is generated using at least two of the
input images 104, one image taken with flash lighting, and a second
image taken without flash lighting. Such images can be taken using,
for example, a digital camera or video camera (also called a
camcorder) that employ the system 100.
[0062] The system 100 also employs a comparison component 112 that
identifies image artifacts by analyzing at least two of the input
images 104. Such artifacts can include red-eye coloration, shadows,
and noise. A cost component 114 applies a cost function to the
frequency components to mitigate combining disagreeing frequency
components. The cost component 114 is part of an algorithm that
first splits the flash/no-flash images into their respective low
and high pass components. Lastly, the images are combined using the
cost function which is expressed as a blending mask.
[0063] Referring now to FIG. 2, there is illustrated a flow chart
of one methodology for new image generation in accordance with the
present invention. While, for purposes of simplicity of
explanation, the one or more methodologies shown herein, e.g., in
the form of a flow chart, are shown and described as a series of
acts, it is to be understood and appreciated that the present
invention is not limited by the order of acts, as some acts may, in
accordance with the present invention, occur in a different order
and/or concurrently with other acts from that shown and described
herein. For example, those skilled in the art will understand and
appreciate that a methodology could alternatively be represented as
a series of interrelated states or events, such as in a state
diagram. Moreover, not all illustrated acts may be required to
implement a methodology in accordance with the present
invention.
[0064] Today, digital photography makes it fast, easy, and
economical to take a pair of images of low-light environments: one
with flash ("flash") to capture detail and one without flash
("no-flash") to capture ambient illumination. At 200, the
flash/no-flash images are taken of the object or scene. At 202,
ambient image denoising is performed. The relatively noise-free
flash image is used to reduce noise in the no-flash image. By
maintaining the natural lighting of the ambient image, the new (or
output) image is created that looks closer to the real scene. At
204, flash-to-ambient detail is transferred. High-frequency detail
is transferred from the flash image to the denoised ambient image,
since this detail may not exist in the original ambient image. At
206, white balancing is performed. The user may desire to simulate
a whiter illuminant while preserving the "feel" of the ambient
image. The known flash color is exploited to white-balance the
ambient image, rather than relying on traditional single-image
heuristics. At 208, continuous flash intensity adjustment can be
optionally provided. Continuous interpolation control is provided
between the image pair so that the user can interactively adjust
the flash intensity. The user can even extrapolate beyond the
original ambient and flash images. At 210, red-eye correction is
performed. Red-eye detection is performed by considering how the
color of the pupil changes between the ambient and flash images.
The process then reaches a Stop block. Thus, the present invention
exploits information of the flash/no-flash pair to improve upon
conventional techniques.
[0065] One feature of the present invention is the manual
acquisition of the flash/no-flash pair that is relatively
straightforward with current consumer digital cameras. In support
thereof, the capability to capture such pairs can move into the
camera firmware, thereby making the acquisition process even easier
and faster.
[0066] BACKGROUND ON CAMERA NOISE
[0067] The intuition behind several disclosed algorithms is that
while the illumination from a flash may change the appearance of
the scene, it also increases the signal-to-noise ratio (SNR) in the
flash image and provides a better estimate of the high-frequency
detail. The digital sensor produces similar log power spectra for
the flash and ambient images. According to the capabilities of one
conventional CCD (Charge Coupled Device) camera, a brighter image
signal contains more noise than a darker signal. However, the
absolute gain in signal power is disproportionately larger than the
increase in noise. Thus, the SNR of the brighter image is better.
While the flash does not illuminate the scene uniformly, it does
significantly increase scene brightness (especially for objects
near the camera) and therefore, the flash image exhibits a better
SNR than the ambient image. With regard to the digital sensor, the
noise dominates the signal at a lower frequency in the high-ISO
(International Standards Organization film sensitivity measure)
ambient image than in the low-ISO flash image.
[0068] The improvement in SNR in a flash image is especially
pronounced at higher frequencies. Properly exposed image pairs have
similar intensities after passing through the imaging system (which
may include aperture, shutter/flash duration, and camera gain, for
example). Therefore, the log power spectra are roughly the same.
However, the noise in the high-ISO ambient image is greater than in
the low-ISO flash image because the gain amplifies the noise. Since
the power spectrum of most natural images falls off at high
frequencies, whereas that of the camera noise remains uniform
(i.e., assuming white noise), noise dominates the signal at a much
lower frequency in the ambient image than in the flash image.
[0069] ACQUISITION
[0070] Referring now to FIG. 3, there is illustrated a flow chart
of one methodology of acquisition in accordance with the present
invention. The disclosed algorithms are designed to work with
images acquired using consumer-grade digital cameras. One goal of
the acquisition procedure is to ensure that the flash/no-flash
image pair capture exactly the same points in the scene. The focal
length and aperture are fixed between the two images so that the
camera focus and depth-of-field remain constant. One implementation
of an acquisition procedure is as follows. At 300, focus is made on
the subject. At 302, the focal length and aperture are locked. At
304, the exposure time t and ISO are set for a good exposure. At
306, the ambient image A is captured. At 308, the flash is turned
on. At 310, the exposure time t and ISO are adjusted to the
smallest settings that still expose the image well. At 312, the
flash image F is then taken. The process then reaches a Stop
block.
[0071] In one implementation for handheld camera operation,
exposure times for a single image are set for under 1/30.sup.th of
a second for a 30 mm lens to prevent motion blur. In another
implementation, exposure times for both images are set to
1/60.sup.th of a second or less, so that under ideal circumstances,
both images could be shot one after another within the 1/30.sup.th
of a second limit for handheld camera operation. Although rapidly
switching between flash and non-flash mode is not currently
possible on consumer grade cameras, it is to be appreciated that
this capability will eventually be included in camera firmware, and
is contemplated as part of the present invention.
[0072] All images can be acquired in RAW digital format. Images can
then be converted into lossless 16-bit TIFF images or lossy JPEG
images. By default, some conventional cameras include conversion
software that performs white balancing, gamma correction, and other
nonlinear tone-mapping operations to produce perceptually pleasing
images with good overall contrast. One or more of the disclosed
algorithms are applied on these non-linear images in order to
preserve their high quality tone-mapping characteristics in the
final output images.
[0073] It is to be appreciated that image registration can be
accomplished using conventional mechanical means such as a tripod
setup. Registration is important for images taken with handheld
cameras, since changing the camera settings (e.g., turning on the
flash, and changing the ISO) often results in camera motion.
Photographs taken without a tripod can also benefit from the
disclosed invention. Image registration from handheld cameras can
be performed algorithmically. Such an algorithm can be found in the
following reference: U.S. Pat. No. 6,018,349 entitled "Patch-Based
Alignment Method and Apparatus for Construction of Image Mosaics,"
by R. Szeliski and H. Shum, which issued Jan. 25, 2000.
[0074] Some of the disclosed algorithms analyze the image
difference F-A to infer the contribution of the flash to the scene
lighting. To make this computation meaningful, the images must be
in the same linear space. Therefore, the conversion software can be
set to generate linear TIFF images from the RAW data. Moreover,
compensation for the exposure differences between the two images
due to ISO settings and exposure times t can be performed. Where
A'.sup.Lin and F.sup.Lin are defined as the linear images output by
the converter utility, they can be put in the same space by
computing: A Lin = A ' .times. .times. Lin .times. ISO F .times.
.DELTA. .times. .times. t F ISO A .times. .DELTA. .times. .times. t
A . ##EQU1##
[0075] Note that unless the superscript Lin is included, F and A
refer to the non-linear versions of the images.
[0076] DENOISING AND DETAIL TRANSFER
[0077] Denoising and detail transfer algorithms are designed to
enhance the ambient image using information from the flash image.
Both algorithms assume that the flash image is a good local
estimator of the high frequency content in the ambient image.
However, this assumption does not hold in shadow and specular
regions caused by the flash, and can lead to artifacts. Artifact
handling is described in greater detail herein below.
[0078] Reducing noise in photographic images has been a
long-standing problem in image processing and computer vision.
Conventional solutions include applying an edge-preserving
smoothing filter to the image such as anisotropic diffusion or
bilateral filtering. The bilateral filter is a fast, non-iterative
technique, and has been applied to a variety of problems beyond
image denoising, including tone mapping, separating illumination
from texture, and mesh smoothing.
[0079] The disclosed ambient image denoising technique also builds
on the bilateral filter.
[0080] Referring now to FIG. 4, there is illustrated a flow chart
of one methodology for denoising in accordance with the present
invention. At 400, the bilateral filter is employed to average
spatially-near pixels having similar intensity values by computing
values of each pixel p for no-flash image A. It combines a classic
low-pass filter with an edge-stopping function that attenuates the
filter kernel weights when the intensity difference between pixels
is large. As noted in the prior art, the bilateral filter computes
the value of pixel p for ambient image A as: A p Base = 1 k
.function. ( p ) .times. p ' .times. .epsilon..OMEGA. .times.
.times. g d .function. ( p ' - p ) .times. g r .function. ( A p - A
p ' ) .times. A p ' , ##EQU2##
[0081] where k(p) is a normalization term: k .function. ( p ) = p '
.times. .epsilon. .times. .times. .OMEGA. .times. .times. g d
.function. ( p ' - p ) .times. g r .function. ( A p - A p ' ) .
##EQU3##
[0082] The function g.sub.d sets the weight in the spatial domain
based on the distance between the pixels, while the edge-stopping
function g.sub.r sets the weight on the range based on intensity
differences. Typically, both functions are Gaussians with widths
controlled by the standard deviation parameters .sigma..sub.d and
.sigma..sub.r, respectively.
[0083] At 402, the bilateral filter is applied to each RGB color
channel separately with the same standard deviation parameters for
all three channels. The challenge is to set .sigma..sub.d and
.sigma..sub.r so that the noise is averaged away but detail is
preserved. In practice, for six megapixel images, .sigma..sub.d is
set to cover a pixel neighborhood of between 24 and 48 pixels, and
then experimentally adjust .sigma..sub.r so that it is just above
the threshold necessary to smooth the noise. For images with pixel
values normalized to [0.0, 1.0], .sigma..sub.r is set to lie
between 0.05 and 0.1, or 5 to 10% of the total range. However, as
shown hereinbelow in FIG. 7b, even after carefully adjusting the
parameters, the basic bilateral filter tends to either over-blur
(lose detail) or under-blur (fail to denoise) the image in some
regions.
[0084] It was observed hereinabove that the flash image contains a
much better estimate of the true high-frequency information than
the ambient image. Based on this observation, the basic bilateral
filter is modified to compute the edge-stopping function g.sub.r
using the flash image F instead of A, which technique called the
joint bilateral filter. At 404, the joint bilateral filter is
employed to average spatially near pixels having similar intensity
values by computing values of each pixel p for the flash image F,
described as follows: A p NR = 1 k .function. ( p ) .times. p '
.times. .epsilon..OMEGA. .times. .times. g d .function. ( p ' - p )
.times. g r .function. ( F p - F p ' ) .times. A p ' , ##EQU4##
[0085] where k(p) is modified similarly. Here A.sup.NR is the
noise-reduced version of A. .sigma..sub.d is set as before for the
basic bilateral filter. Under the assumption that F has little
noise, .sigma..sub.r can be set to be very small and still ensure
that the edge-stopping function g.sub.r(F.sub.p-F.sub.p'), will
choose the proper weights for nearby pixels, and therefore, will
not over-blur or under-blur the ambient image. In practice,
.sigma..sub.r can be set to 0.1% of the total range of color
values. Unlike basic bilateral filtering, .sigma..sub.r is fixed
for all images.
[0086] The joint bilateral filter relies on the flash image as an
estimator of the ambient image. Therefore, it can fail in flash
shadows and specularities because they only appear in the flash
image. At the edges of such regions, the joint bilateral filter may
under-blur the ambient image since it will down-weight pixels where
the filter straddles these edges. Similarly, inside these regions,
it may over-blur the ambient image. At 406, this problem is solved
by first detecting flash shadows and specular regions, and then
falling back to basic bilateral filtering within these regions, as
indicated at 408. The process then reaches a Stop block.
[0087] Given the mask M produced by our detection algorithm, our
improved denoising algorithm becomes:
A.sup.NR'=(1-M)A.sup.NR+MA.sup.BASE.
[0088] The results of denoising with the joint bilateral filter are
shown in FIG. 7c. The difference image with the basic bilateral
filter in FIG. 7d, reveals that the joint bilateral filter is
better able to preserve detail while reducing noise. Since both
bilateral and joint bilateral filtering is nonlinear, a
straightforward implementation requires performing the convolution
in the spatial domain. This can be very slow for large
.sigma..sub.d. One conventional implementation for accelerating the
denoising algorithm uses Fourier techniques. This technique is also
applicable to the joint bilateral filter and can significantly
speed up the disclosed denoising algorithm.
[0089] Referring now to FIG. 5, there is illustrated a flow chart
of one methodology for detail transfer in accordance with the
present invention. While the joint bilateral filter can reduce
noise, it cannot add detail that may be present in the flash image.
Yet, as described hereinabove, the higher SNR of the flash image
allows it to retain nuances that are overwhelmed by noise in the
ambient image. Moreover, the flash typically provides strong
directional lighting that can reveal additional surface detail that
is not visible in more uniform ambient lighting. The flash may also
illuminate detail in regions that are in shadows in the ambient
image. To transfer this detail, at 500, a detail layer is computed
from the flash image as the following ratio: F Detail = F + F Base
+ , ##EQU5##
[0090] where F.sup.Base is computed using the basic bilateral
filter on F. The ratio is computed on each RGB channel separately
and is independent of the signal magnitude and surface reflectance,
as indicated at 502. The ratio captures the local detail variation
in F and is commonly called a quotient image or ratio image in
computer vision. An advantage of using the bilateral filter to
compute F.sup.Base rather than a classic low-pass Gaussian filter
is for reducing haloing. A Gaussian low-pass filter blurs across
all edges and will therefore create strong peaks and valleys in the
detail image that cause halos. The bilateral filter does not smooth
across strong edges and thereby reduces halos, while still
capturing detail.
[0091] At low signal values, the flash image contains noise that
can generate spurious detail. At 504, the algorithm compensates for
low signal values by adding value .epsilon. to both the numerator
and denominator of the ratio to reject these low signal values, and
thereby reduce such artifacts (and also avoid division by zero). In
practice .epsilon.=0.02 is used across all results. To transfer the
detail, the noise-reduced ambient image A.sup.NR is multiplied by
the ratio F.sup.Detail. FIG. 7e and FIG. 7f show examples of a
detail layer and detail transfer.
[0092] Just as in joint bilateral filtering, the transfer algorithm
can produce a poor detail estimate in shadows and specular regions
caused by the flash. Therefore, at 506, the detection algorithm is
applied to estimate a mask M identifying these regions and compute
the final image as:
A.sup.Final=(1-M)A.sup.NRF.sup.Detail+MA.sup.Base.
[0093] With this detail transfer approach, the amount of detail
transferred can be controlled by choosing appropriate settings for
the bilateral filter parameters .sigma..sub.d and .sigma..sub.r
used to create F.sup.Base. As the filter widths increase,
increasingly smoother versions of F.sup.Base are generated and, as
a result, capture more detail in F.sup.Detail. However, with
excessive smoothing, the bilateral filter essentially reduces to a
Gaussian filter and leads to haloing artifacts in the final
image.
[0094] Depending on the scene, the extreme levels of noise can
require the use of relatively wide Gaussians for both the domain
and range kernels in the joint bilateral filter. Thus, when
transferring back the true detail from the flash image, a
relatively wide Gaussians was used in computing the detail layer.
As a result, it is possible to see small halos around the edges of
the bottles illustrated in FIG. 9b. Nevertheless, the disclosed
architecture is able to smooth away the noise while preserving
detail.
[0095] In most cases, the detail transfer algorithm improves the
appearance of the ambient image. However, it is important to note
that the flash image may contain detail that looks unnatural when
transferred to the ambient image. For example, if the light from
the flash strikes a surface at a shallow angle, the flash image may
pick up surface texture (e.g., wood grain, and stucco) as detail.
If this texture is not visible in the original ambient image, it
may look odd. Similarly if the flash image washes out detail, the
ambient image may be over-blurred. The disclosed algorithms allow
the user to control how much detail is transferred over the entire
image. In another implementation, the amount of local detail
transferred is automatically adjusted.
[0096] Referring now to FIG. 6, there is illustrated an overview of
a denoising algorithm 600, a detail transfer algorithm 602, and
flash artifact detection algorithm 604 that operate on a no-flash
image 606 and a flash image 608, in accordance with the present
invention. The ambient (or no-flash) image 606 is processed using
both a first bilateral filter 610 and a joint bilateral filter 612,
the respective outputs of which are A.sup.BASE and A.sup.NR.
Another input to the joint bilateral filter 612 is the flash image
608. The joint bilateral filter 612 receives these inputs and
outputs the A.sup.NR value to a product process 614.
[0097] The detail transfer algorithm 602 uses the flash image 608
as an input to a second bilateral filter 616, the output of which
is F.sup.BASE. A division process 618 takes as inputs the
F.sup.BASE value and the flash image 608, and outputs the
A.sup.DETAIL value to the product process 614.
[0098] The artifact detection algorithm 604 includes a shadow and
specularity detection algorithm 620 that receives as input
F.sup.LIN and A.sup.LIN from the flash image 608, since the flash
image 608 interfaces to the no-flash image 606. An output of the
shadow and specularity detection algorithm is the Mask M.
[0099] Mask M, A.sup.BASE and, the product of A.sup.NR and
A.sup.DETAIL combine to form the final image A.sup.FINAL.
[0100] Referring now to FIGS. 7a-f, there are illustrated process
shots generated from flash/no-flash images in accordance with the
present invention. FIG. 7a is close-up of a flash image 700 and a
no-flash image 702 of a Belgian tapestry. The no-flash image 702 is
especially noisy in the darker regions and does not show the
threads as well as the flash image 700. FIG. 7b shows a denoised
image via basic bilateral filtering to preserve strong edges, but
blurs away most of the threads. FIG. 7c shows a denoised image
using joint bilateral filtering. Joint bilateral filtering smoothes
the noise while also retaining more thread detail than the basic
bilateral filter. FIG. 7d shows the difference image between the
basic and joint bilateral filtered images. FIG. 7e show the
generated detail layer. The ambient image is further enhanced by
transferring detail from the flash image. The detail layer is first
computed from the flash image, and then combined with the image
denoised via the joint bilateral filter to produce the
detail-transferred image, as illustrated in FIG. 7f. The difference
image with the basic bilateral filter of FIG. 7d reveals that the
joint bilateral filter is better able to preserve detail while
reducing noise.
[0101] Referring now to FIGS. 8a-e, there are illustrated shots of
another example that employs detail transfer and denoising to
maintain the original warm appearance in accordance with the
present invention. In FIG. 8a, a flash image and a no-flash image
are provided of an old European lamp made of hay. The flash image
captures detail, but is gray and flat. The no-flash image captures
the warm illumination of the lamp, but is noisy and lacks the fine
detail of the hay. FIG. 8b illustrates a small section of the image
that is examined for final results. FIG. 8c shows the flash version
of the section. FIG. 8d shows the no-flash version of the section.
FIG. 8e shows the detail transfer with denoising, that maintains
the warm appearance, as well as the sharp detail.
[0102] Referring now to FIGS. 9a-c, there are illustrated shots of
a long exposure reference to the detail transfer and denoising
result of FIG. 8. FIG. 9a shows a no-flash image. FIG. 9b shows the
detail transfer with denoising. FIG. 9c shows a long exposure image
of the wine cave scene (3.2 seconds at ISO 100) that is captured
for comparison with detail transfer with denoising result of FIG.
9b. Visual comparison shows that although the detail transfer
result does not achieve the fidelity of the reference image, it is
substantially less noisy than the original no-flash image.
[0103] DETECTING FLASH SHADOWS AND SPECULARITIES
[0104] Light from the flash can introduce shadows and specularities
into the flash image. Within flash shadows, the image may be as dim
as the ambient image and therefore suffer from noise. Similarly,
within specular reflections, the flash image may be saturated and
lose detail. Moreover, the boundaries of both these regions may
form high-frequency edges that do not exist in the ambient image.
To avoid using information from the flash image in these regions,
the flash shadows and specularities are first detected.
[0105] Referring now to FIG. 10, there is illustrated a flow chart
of one methodology for flash shadow and specularity detection in
accordance with the present invention. Since a point in a flash
shadow is not illuminated by the flash, it should appear exactly as
it appears in the ambient image. Ideally, A and F can be linearized
as described hereinabove, and then pixels detected where the
luminance of the difference image F.sup.Lin-A.sup.Lin is zero. In
practice, this approach is confounded by four issues: 1) surfaces
that do not reflect any light (i.e., with zero albedo) are detected
as shadows; 2) distant surfaces not reached by the flash are
detected as shadows; 3) noise causes nonzero values within shadows;
and 4) inter-reflection of light from the flash causes non-zero
values within the shadow.
[0106] At 1000, zero albedo surfaces and shadows are addressed. The
first two issues do not cause a problem since the results are the
same in both the ambient and flash images, and thus, whichever
image is chosen will give the same result. At 1002, noise and
inter-reflection are addressed by adding a threshold when computing
the shadow mask by looking for pixels in which the difference
between the linearized flash and ambient images is small, according
to the flowing conditions: M Shad = { 1 , when .times. .times. F
Lin - A Lin .ltoreq. .tau. Shad 0 .times. .times. else .
##EQU6##
[0107] One of the disclosed algorithms lets users interactively
adjust the threshold value .tau..sub.Shad and visually verify that
all the flash shadow regions are properly captured, as indicated at
1004.
[0108] Noise can contaminate the shadow mask with small speckles,
holes and ragged edges. The shadow mask is cleaned up using image
morphological operations to erode the speckles and fill the holes,
as indicated at 1006. At 1008, to produce a conservative estimate
that fully covers the shadow region, the mask is then dilated.
[0109] At 1010, specular regions caused by the flash are detected
using a simple physically motivated heuristic. Specular regions
should be bright in Lin F and should therefore saturate the image
sensor. Hence, luminance values in the flash image that are greater
than 95% of the range of sensor output values are sought. At 1012,
cleaning, hole filling, and dilating of the specular mask are
performed as before for the shadow mask. At 1014, the final mask M
is formed by taking the union of the shadow and specular masks. At
1016, the mask is the blurred to feather its edges and prevent
visible seams when the mask is used to combine regions from
different images.
[0110] Referring now to FIG. 11, there are illustrated image shots
for artifact processing in accordance with the present invention.
FIG. 11a illustrates flash and no-flash images. FIG. 11b shows the
detail transfer image for the no-flash image of FIG. 11a. FIG. 11c
shows the detail transfer without Mask of the section of the flash
image, where shadows are outlined at the arrows. FIG. 11d shows the
shadow and specularity mask. FIG. 11e shows the detail transfer
using the Mask of the same image of FIG. 11c. FIG. 11f shows a
flash image of the small section indicated in the no-flash image of
FIG. 11a. FIG. 11g shows the no-flash version of the same small
section of FIG. 11f. FIG. 11h shows the detail transfer with
denoising of the images of FIG. 11f and FIG. 11g.
[0111] The flash image does not contain true detail information in
shadows and specular regions. When naively applying the denoising
and detail transfer algorithms, these regions generate artifacts,
as indicated by the white arrows. To prevent these artifacts, basic
bilateral filtering is employed within these regions. The dark
brown pot on the left in the no-flash image of FIG. 11a is
extremely noisy. The green pot on the right of FIG. 11a is also
noisy, but as shown in the flash image of FIG. 11a, exhibits true
texture detail. The detail transfer technique smoothes the noise
while maintaining the texture, as shown in FIG. 11e. Note that the
flash shadow/specularity detection algorithm properly masks out the
large specular highlight on the brown pot of FIG. 11d and does not
transfer that detail to the final image of FIG. 11e.
[0112] WHITE BALANCING
[0113] Although preserving the original ambient illumination is
often desirable, sometimes it is also desirable to see how the
scene would appear under a more "white" illuminant, in a process is
called white-balancing.
[0114] When only a single ambient image is acquired, the ambient
illumination must be estimated based on heuristics or user input.
Digital cameras usually provide several white-balance modes for
different environments such as sunny outdoors and fluorescent
lighting. Most often, pictures are taken with an "auto" mode,
wherein the camera analyzes the image and computes an image-wide
average to infer ambient color. This is, of course, only a
heuristic, and some researchers have considered semantic analysis
to determine color cast.
[0115] A flash/no-flash image pair enables a better approach to
white balancing. The disclosed architecture requires less setup
than conventional processes by formulating white balancing as a
continuous optimization problem that is not limited by the
conventional discrete set of illuminants.
[0116] Referring now to FIG. 12, there is illustrated a flow chart
of one methodology for white balancing in accordance with the
present invention. At 1200, the no-flash image is generated. At
1202, the white-balancing mode of the camera is set to flash. A
flash can be considered as adding a point light source of known
color to the scene. By setting the camera white-balance mode to
"flash" (and assuming a calibrated camera), this flash color should
appear as reference white in the acquired images. At 1204, the
flash image is generated.
[0117] At 1206, the difference image is computed. The difference
image .DELTA.=F.sup.Lin-A.sup.Lin corresponds to the illumination
due to the flash only, which is proportional to the surface albedo
at each pixel p. Note that the albedo estimate .DELTA. has unknown
scale, because both the distance and orientation of the surface are
unknown. It is assumed either that the surface is diffuse or that
its specular color matches its diffuse color. As a counter-example,
this is not true of plastics. Similarly, semitransparent surfaces
would give erroneous estimates of albedo.
[0118] At 1208, the ambient illumination for each color channel is
estimated. Since the surface at pixel p has color A.sub.p in the
ambient image and the scaled albedo .DELTA..sub.p, the ambient
illumination at the surface can be estimated with the following
ratio: C p = .DELTA. p A p , ##EQU7##
[0119] which is computed per color channel. Again, this estimated
color C.sub.p has an unknown scale, so it is normalized at each
pixel p. A goal is to analyze C.sub.p at all image pixels to infer
the ambient illumination color c. To make this inference more
robust, pixels for which the estimate has low confidence are
discarded. This can be done since only a single color need be
derived from millions of pixels. Specifically, pixels are ignored
for which either |A.sub.p<.tau..sub.1 or the luminance of
.DELTA..sub.p<.tau..sub.2 in any channel, since these small
values make the ratio less reliable. Both .tau..sub.1 and
.tau..sub.2 are set to about 2% of the range of color values.
[0120] Finally, at 1210, the ambient color estimate c for the scene
is computed as the mean of C.sub.p for the non-discarded pixels. An
alternative is to select c as the principal component of C,
obtained as the eigenvector of C.sup.TC with the largest
eigenvalue, and this gives a similar answer. Having inferred the
scene ambient color c, the image is white-balanced, at 1212, by
scaling the color channels as: A p WB = 1 c .times. A p .
##EQU8##
[0121] Again, the computation is performed per color channel.
[0122] Referring now to FIGS. 13a-c, there are illustrated image
shots associated with white-balancing an ambience image in
accordance with the present invention. FIG. 13a shows an original
no-flash image after denoising and detail transfer, but which still
shows a cast. FIG. 13b shows the estimated ambient illumination
colors and the estimated overall scene ambience. FIG. 13c shows
that the white-balancing algorithm shifts the colors and removes a
certain coloring (e.g., orange).
[0123] The white balancing significantly changes the overall hue of
the image, setting the color of the wood table to a yellowish gray,
as it would appear in white light. In inferring ambient color c,
one could also prune outliers and look for spatial relationships in
the image C. In addition, the scene may have multiple regions with
different ambient colors, and these could be segmented and
processed independently. White-balancing is a challenging problem
because the perception of "white" depends in part on the adaptation
state of the viewer. Moreover, it is unclear when white-balance is
desirable. However, the disclosed estimation approach using the
known information from the flash can be more accurate than
techniques based on single-image heuristics.
[0124] CONTINUOUS FLASH ADJUSTMENT
[0125] When taking a flash image, the intensity of the flash can
sometimes be too bright, saturating a nearby object, or it can be
too dim, leaving mid-distance objects under-exposed. With a flash
and non-flash image pair, the present invention allows the user to
adjust the flash intensity after the picture has been taken.
[0126] Referring now to FIG. 14, there is illustrated a flow chart
of one methodology for continuous flash adjustment in accordance
with the present invention. At 1400, the flash and no-flash images
are generated. At 1402, the flash and no-flash images are
interpolated. One way of interpolating the ambient and flash images
is to convert the original flash/no-flash pair into YCbCr space,
and then linearly interpolate them using:
F.sup.Adjusted=(1-.alpha.)A+(.alpha.)F.
[0127] To provide more user control, extrapolation is allowed by
letting the parameter .alpha. go outside the normal [0,1] range.
However, only the Y channel is extrapolated, and the Cb and Cr
channel interpolations are restricted to their extrema in the two
original images to prevent excessive distortion of the hue. An
example is shown in FIG. 15.
[0128] FIGS. 15a-f illustrate an example of continuous flash
adjustment by extrapolation between flash and no-flash images. FIG.
15a shows an out-of-range extreme at the low end with a setting at
-0.5. FIG. 15b shows the no-flash image at 0.0. FIG. 15c shows an
extrapolated image with a 0.33 setting. FIG. 15d shows an
extrapolated image with a 0.66 setting. FIG. 15e shows the flash
image at the 1.0 setting. FIG. 15f shows an extrapolated image with
an out-of-range extreme at the high end at 1.5.
[0129] RED-EYE CORRECTION
[0130] Red-eye is a common problem in flash photography and is due
to light reflected by a well vascularized retina. Fully automated
redeye removal techniques conventionally assume a single image as
input and rely on a variety of heuristic and machine-learning
techniques to localize the red eyes. Once the pupil mask has been
detected, these techniques darken the pixels within the mask to
make the images appear more natural.
[0131] Referring now to FIG. 16, there is illustrated a red-eye
removal methodology in accordance with the present invention. The
red-eye removal algorithm of the present invention considers the
change in pupil color between the ambient image (where it is
usually very dark) and the flash image (where it may be red). At
1600, the image pair is converted into YCbCr space to decorrelate
luminance from chrominance. At 1602, a relative redness measure is
computed, as follows: R=F.sub.Cr-A.sub.Cr.
[0132] At 1604, the image is initially segmented into regions
where: R>.tau..sub.Eye.
[0133] The parameter .tau..sub.Eye is typically set to 0.05 so that
the resulting segmentation defines regions where the flash image is
more red than the ambient image, and therefore, may form potential
red eyes. The segmented regions also tend to include a few
locations that are highly saturated in the Cr channel of the flash
image but are relatively dark in the Y channel of the ambient
image. Thus, if .mu..sub.R and .sigma..sub.R denote the mean and
standard deviation of the redness R, seed pixels are searched, as
indicated at 1606, where: R>max[0.6, .mu..sub.R+3.sigma..sub.R]
and A.sub..gamma.<.tau..sub.Dark, and where .tau..sub.Dark is
typically set to 0.6.
[0134] At 1608, if no such seed pixels exist, it can be assumed
that the image does not contain red-eye, and flow is to a Stop
block. Otherwise, flow is to 1610, where the seed pixels are used
to find the corresponding regions in the segmentation. At 1612,
geometric constraints are applied to ensure that the regions are
roughly the same size and elliptical. In particular, the area of
each region is computed and large outliers discarded. At 1614, the
eccentricity of the region is checked to ensure that it is greater
than 0.75, which regions form a red-eye pupil mask. The red-eye
regions are removed by first removing the highlights or "glints" in
the pupil mask using the previously described flash specularity
detection algorithm, as indicated at 1616. At 1618, the color of
each pixel in the mask is set to the gray value equivalent to 80%
of its luminance value. This approach properly darkens the pupil
while maintaining the specular highlight which is important for
maintaining realism in the corrected output. The process then
reaches the Stop block.
[0135] In another implementation, an infrared flash can be
employed. While infrared illumination yields incomplete color
information, it does provide high-frequency detail, and does so in
a less intrusive way than a visible flash.
[0136] Referring now to FIG. 17, there is illustrated a block
diagram of a small form factor portable device 1700 that includes
the image processing architecture of the present invention. The
device 1700 includes a processor 1702 for controlling all onboard
operations and processes. A memory 1704 interfaces to the processor
1702 for temporary storage of data and one or more device
applications 1706 for image processing in accordance with the
present invention being executed by the processor 1702.
[0137] A communications component 1708 interfaces to the processor
1702 to facilitate wired/wireless communication with suitable
external systems. This can include IEEE 802.11-based wireless
communications and telecommunications signals based on conventional
air protocols for mobile telephone signals.
[0138] The device 1700 can include a display 1710 for presenting at
least image content captured in accordance with the present
invention. The display 1710 can also facilitate the presentation of
setup and configuration information for operating the device 1700
in the form of text and or graphics for using the device features.
A serial I/O interface 1712 is provided in communication with the
processor 1702 to facilitate serial communication (e.g., USB,
and/or IEEE 1394) via a hardwire connection. This supports updating
and troubleshooting, and uploading/downloading image data to/from
the device 1700, for example. Audio capabilities are provided with
an audio I/O component 1714, which can include a speaker for the
output of audio signals related to, for example, recorded data or
telephony voice data, and a microphone for inputting voice signals
for recording and/or telephone conversations.
[0139] The device 1700 can include firmware 1716 to provide
non-volatile storage and and access to the processor 1702 of
startup and operation instructions.
[0140] The device 1700 can also include an image capture subsystem
1718 that includes an image capture subsystem such as a CCD (Charge
Coupled Device) or CMOS (Complementary Metal Oxide Semiconductor)
imager, for example. The image capture subsystem 1718 includes
suitable memory and buffering capability to support rapid
successive imaging of an object and/or scene for both flash and
no-flash imaging and processing in accordance with the present
invention. In one implementation, the subsystem 1718 can include
the capability to provide flash and no-flash images by interleaved
scanning of the object or scene. That is, the flash image is
captured by the odd pixel lines and the no-flash image is captured
according to the even pixel lines of the imager. In another
implementation, the imaging subsystem comprises twin capture
subsystems such that the flash image is captured by a first
subsystem and the no-flash image is captured a second subsystem. In
any case, the flash/no-flash images are captured in rapid
succession of one another. In yet another implementation, a single
buffer is provided that captures successive images taken in rapid
succession by the imaging subsystem that takes the ambient image
and then the flash image, or vice versa, with suitable time
therebetween to provide the desired flash/no-flash images.
[0141] The display 1710 can be a pixel-based display (e.g., an LCD)
such that the imaging subsystem 1718 supports a double pixel
buffered CCD/CMOS design for processing multiple images in
accordance with the present invention.
[0142] Given that the device 1700 include the communications
component 1708, a user can then take digital pictures, and transmit
the pictures to a remote location or store the pictures locally.
The device 1700 can also include a power source 1720 in the form of
batteries, which power source 1720 can also interface to an
external power system or charging equipment via a power I/O
component 1722.
[0143] The device software 1706 can also include one or more
programs such as an operating system for configuring and
manipulating local data and settings, a browser for interacting
with websites, music players, video player software, and any other
software suitable for operation on the device 1000.
[0144] The device 1700 can be a digital camera, an MP3 player with
image capture capability, a cellular telephone with a built-in
digital camera, a PDA (person digital assistant), or any such
portable device suitably designed to accommodate multi-image
processing in accordance with the present invention.
[0145] Referring now to FIG. 18, there is illustrated a block
diagram of a computer operable to execute the disclosed
architecture. In order to provide additional context for various
aspects of the present invention, FIG. 18 and the following
discussion are intended to provide a brief, general description of
a suitable computing environment 1800 in which the various aspects
of the present invention can be implemented. While the invention
has been described above in the general context of
computer-executable instructions that may run on one or more
computers, those skilled in the art will recognize that the
invention also can be implemented in combination with other program
modules and/or as a combination of hardware and software.
[0146] Generally, program modules include routines, programs,
components, data structures, etc., that perform particular tasks or
implement particular abstract data types. Moreover, those skilled
in the art will appreciate that the inventive methods can be
practiced with other computer system configurations, including
single-processor or multiprocessor computer systems, minicomputers,
mainframe computers, as well as personal computers, hand-held
computing devices, microprocessor-based or programmable consumer
electronics, and the like, each of which can be operatively coupled
to one or more associated devices.
[0147] The illustrated aspects of the invention may also be
practiced in distributed computing environments where certain tasks
are performed by remote processing devices that are linked through
a communications network. In a distributed computing environment,
program modules can be located in both local and remote memory
storage devices.
[0148] A computer typically includes a variety of computer-readable
media. Computer-readable media can be any available media that can
be accessed by the computer and includes both volatile and
nonvolatile media, removable and non-removable media. By way of
example, and not limitation, computer readable media can comprise
computer storage media and communication media. Computer storage
media includes both volatile and nonvolatile, removable and
non-removable media implemented in any method or technology for
storage of information such as computer readable instructions, data
structures, program modules or other data. Computer storage media
includes, but is not limited to, RAM, ROM, EEPROM, flash memory or
other memory technology, CD-ROM, digital video disk (DVD) or other
optical disk storage, magnetic cassettes, magnetic tape, magnetic
disk storage or other magnetic storage devices, or any other medium
which can be used to store the desired information and which can be
accessed by the computer.
[0149] Communication media typically embodies computer-readable
instructions, data structures, program modules or other data in a
modulated data signal such as a carrier wave or other transport
mechanism, and includes any information delivery media. The term
"modulated data signal" means a signal that has one or more of its
characteristics set or changed in such a manner as to encode
information in the signal. By way of example, and not limitation,
communication media includes wired media such as a wired network or
direct-wired connection, and wireless media such as acoustic, RF,
infrared and other wireless media. Combinations of the any of the
above should also be included within the scope of computer-readable
media.
[0150] With reference again to FIG. 18, there is illustrated an
exemplary environment 1800 for implementing various aspects of the
invention that includes a computer 1802, the computer 1802
including a processing unit 1804, a system memory 1806 and a system
bus 1808. The system bus 1808 couples system components including,
but not limited to, the system memory 1806 to the processing unit
1804. The processing unit 1804 can be any of various commercially
available processors. Dual microprocessors and other
multi-processor architectures may also be employed as the
processing unit 1804.
[0151] The system bus 1808 can be any of several types of bus
structure that may further interconnect to a memory bus (with or
without a memory controller), a peripheral bus, and a local bus
using any of a variety of commercially available bus architectures.
The system memory 1806 includes read only memory (ROM) 1810 and
random access memory (RAM) 1812. A basic input/output system (BIOS)
is stored in a non-volatile memory 1810 such as ROM, EPROM, EEPROM,
which BIOS contains the basic routines that help to transfer
information between elements within the computer 1802, such as
during start-up. The RAM 1812 can also include a high-speed RAM
such as static RAM for caching data.
[0152] The computer 1802 further includes an internal hard disk
drive (HDD) 1814 (e.g., EIDE, SATA), which internal hard disk drive
1814 may also be configured for external use in a suitable chassis
(not shown), a magnetic floppy disk drive (FDD) 1816, (e.g., to
read from or write to a removable diskette 1818) and an optical
disk drive 1820, (e.g., reading a CD-ROM disk 1822 or, to read from
or write to other high capacity optical media such as the DVD). The
hard disk drive 1814, magnetic disk drive 1816 and optical disk
drive 1820 can be connected to the system bus 1808 by a hard disk
drive interface 1824, a magnetic disk drive interface 1826 and an
optical drive interface 1828, respectively. The interface 1824 for
external drive implementations includes at least one or both of
Universal Serial Bus (USB) and IEEE 1394 interface
technologies.
[0153] The drives and their associated computer-readable media
provide nonvolatile storage of data, data structures,
computer-executable instructions, and so forth. For the computer
1802, the drives and media accommodate the storage of any data in a
suitable digital format. Although the description of
computer-readable media above refers to a HDD, a removable magnetic
diskette, and a removable optical media such as a CD or DVD, it
should be appreciated by those skilled in the art that other types
of media which are readable by a computer, such as zip drives,
magnetic cassettes, flash memory cards, cartridges, and the like,
may also be used in the exemplary operating environment, and
further, that any such media may contain computer-executable
instructions for performing the methods of the present
invention.
[0154] A number of program modules can be stored in the drives and
RAM 1812, including an operating system 1830, one or more
application programs 1832, other program modules 1834 and program
data 1836. All or portions of the operating system, applications,
modules, and/or data can also be cached in the RAM 1812.
[0155] It is appreciated that the present invention can be
implemented with various commercially available operating systems
or combinations of operating systems.
[0156] A user can enter commands and information into the computer
1802 through one or more wired/wireless input devices, e.g., a
keyboard 1838 and a pointing device, such as a mouse 1840. Other
input devices (not shown) may include a microphone, an IR remote
control, a joystick, a game pad, a stylus pen, touch screen, or the
like. These and other input devices are often connected to the
processing unit 1804 through an input device interface 1842 that is
coupled to the system bus 1808, but can be connected by other
interfaces, such as a parallel port, an IEEE 1394 serial port, a
game port, a USB port, an IR interface, etc.
[0157] A monitor 1844 or other type of display device is also
connected to the system bus 1808 via an interface, such as a video
adapter 1846. In addition to the monitor 1844, a computer typically
includes other peripheral output devices (not shown), such as
speakers, printers etc.
[0158] The computer 1802 may operate in a networked environment
using logical connections via wired and/or wireless communications
to one or more remote computers, such as a remote computer(s) 1848.
The remote computer(s) 1848 can be a workstation, a server
computer, a router, a personal computer, portable computer,
microprocessor-based entertainment appliance, a peer device or
other common network node, and typically includes many or all of
the elements described relative to the computer 1802, although, for
purposes of brevity, only a memory storage device 1850 is
illustrated. The logical connections depicted include
wired/wireless connectivity to a local area network (LAN) 1852
and/or larger networks, e.g., a wide area network (WAN) 1854. Such
LAN and WAN networking environments are commonplace in offices, and
companies, and facilitate enterprise-wide computer networks, such
as intranets, all of which may connect to a global communication
network, e.g., the Internet.
[0159] When used in a LAN networking environment, the computer 1802
is connected to the local network 1852 through a wired and/or
wireless communication network interface or adapter 1856. The
adaptor 1856 may facilitate wired or wireless communication to the
LAN 1852, which may also include a wireless access point disposed
thereon for communicating with the wireless adaptor 1856. When used
in a WAN networking environment, the computer 1802 can include a
modem 1858, or is connected to a communications server on the LAN,
or has other means for establishing communications over the WAN
1854, such as by way of the Internet. The modem 1858, which can be
internal or external and a wired or wireless device, is connected
to the system bus 1808 via the serial port interface 1842. In a
networked environment, program modules depicted relative to the
computer 1802, or portions thereof, can be stored in the remote
memory/storage device 1850. It will be appreciated that the network
connections shown are exemplary and other means of establishing a
communications link between the computers can be used.
[0160] The computer 1802 is operable to communicate with any
wireless devices or entities operatively disposed in wireless
communication, e.g., a printer, scanner, desktop and/or portable
computer, portable data assistant, communications satellite, any
piece of equipment or location associated with a wirelessly
detectable tag (e.g., a kiosk, news stand, restroom), and
telephone. This includes at least Wi-Fi and Bluetooth.TM. wireless
technologies. Thus, the communication can be a predefined structure
as with conventional network or simply an ad hoc communication
between at least two devices.
[0161] Wi-Fi, or Wireless Fidelity, allows connection to the
Internet from a couch at home, a bed in a hotel room or a
conference room at work, without wires. Wi-Fi is a wireless
technology like a cell phone that enables such devices, e.g.,
computers, to send and receive data indoors and out; anywhere
within the range of a base station. Wi-Fi networks use radio
technologies called IEEE 802.11 (a, b, g, etc.) to provide secure,
reliable, fast wireless connectivity. A Wi-Fi network can be used
to connect computers to each other, to the Internet, and to wired
networks (which use IEEE 802.3 or Ethernet). Wi-Fi networks operate
in the unlicensed 2.4 and 5 GHz radio bands, with an 11 Mbps
(802.11b) or 54 Mbps (802.11a) data rate or with products that
contain both bands (dual band), so the networks can provide
real-world performance similar to the basic 10BaseT wired Ethernet
networks used in many offices.
[0162] Referring now to FIG. 19, there is illustrated a schematic
block diagram of an exemplary computing environment 1900 in
accordance with the present invention. The system 1900 includes one
or more client(s) 1902. The client(s) 1902 can be hardware and/or
software (e.g., threads, processes, computing devices). The
client(s) 1902 can house cookie(s) and/or associated contextual
information by employing the present invention, for example. The
system 1900 also includes one or more server(s) 1904. The server(s)
1904 can also be hardware and/or software (e.g., threads,
processes, computing devices). The servers 1904 can house threads
to perform transformations by employing the present invention, for
example. One possible communication between a client 1902 and a
server 1904 can be in the form of a data packet adapted to be
transmitted between two or more computer processes. The data packet
may include a cookie and/or associated contextual information, for
example. The system 1900 includes a communication framework 1906
(e.g.,a global communication network such as the Internet) that can
be employed to facilitate communications between the client(s) 1902
and the server(s) 1904.
[0163] Communications can be facilitated via a wired (including
optical fiber) and/or wireless technology. The client(s) 1902 are
operatively connected to one or more client data store(s) 1908 that
can be employed to store information local to the client(s) 1902
(e.g., cookie(s) and/or associated contextual information).
Similarly, the server(s) 1904 are operatively connected to one or
more server data store(s) 1910 that can be employed to store
information local to the servers 1904.
[0164] What has been described above includes examples of the
present invention. It is, of course, not possible to describe every
conceivable combination of components or methodologies for purposes
of describing the present invention, but one of ordinary skill in
the art may recognize that many further combinations and
permutations of the present invention are possible. Accordingly,
the present invention is intended to embrace all such alterations,
modifications and variations that fall within the spirit and scope
of the appended claims. Furthermore, to the extent that the term
"includes" is used in either the detailed description or the
claims, such term is intended to be inclusive in a manner similar
to the term "comprising" as "comprising" is interpreted when
employed as a transitional word in a claim.
* * * * *